132 results about "Pattern recognition" patented technology

A computer-implemented method for tracking a human face in a target video includes obtaining target video data of a human face; and estimating parameters of a target human face model, based on the target video data. A first subset of the parameters represents a geometric shape and a second subset of the parameters represents an expression of the human face. At least one of the estimated parameters is modified in order to obtain new parameters of the target human face model, and output video data are generated based on the new parameters of the target human face model and the target video data.

Systems, methods, and apparatus for neuro-robotic tracking point selection are disclosed. A described example method to control a robot arm includes presenting one or more potential trackable features of a target object, emphasizing at least one of the potential trackable features, determining a selection of one of the emphasized features as a tracking point by monitoring a first mental response to emphasizing the feature, and effectuating an end goal via a robot by tracking the selected feature and effectuating a position based on the selected feature.

The invention is suitable for the technical field of communication, and provides an identity recognition method. The method comprises the following steps of: storing the original fingerprint information of a user; randomly sending a random fingerprint verification instruction after receiving an identity recognition request of the user; acquiring fingerprint information input by the user according to the random fingerprint verification instruction; and judging whether the fingerprint information input by the user is matched with the random fingerprint verification instruction and the original fingerprint information simultaneously, if so, indicating that identity recognition is successful, and otherwise, indicating that identity recognition fails. Correspondingly, the invention also provides an identity recognition system and an identity recognition terminal. Therefore, a safer and more reliable identity recognition technology can be provided, and the implementation mode is simpler.

A gesture recognition method, apparatus, device and operation method based on gesture recognition are disclosed. The gesture recognition method comprises obtaining a corresponding signal characteristic parameter to an action of a current user; and recognizing a corresponding gesture action to the signal characteristic parameter according to a gesture characteristic parameter module matching the identity information of the current user. The gesture recognition method, apparatus, device and operation method based on the gesture recognition can improve the recognition accuracy and versatility of the gesture recognition method.

A computing system for automatically identifying individual regions in a digital image that include a human face, forming combined padded regions and automatically evaluating each of the regions including calculating a fitness score for each of the combined padded regions and ignoring at least one of the combined padded regions whose fitness score is below a known threshold.

The invention is applicable to the technical field of cloud computing, and provides a cloud computing application method, a cloud computing application system and terminal equipment, and a cloud computing platform. The cloud computing application method comprises the following steps that: the terminal equipment acquires and recognizes a face image and then transmits the recognition results; and the cloud computing platform receives the recognition results and implements the application related to the recognition results according to the recognition results. The cloud computing application method provided by the invention has the benefits that cloud computing application is combined with a face recognition technology, the face image is acquired and recognized by utilizing the terminal equipment, and the corresponding application is implemented by the cloud computing platform according to the face image acquired by the terminal equipment; the cloud computing application method can be widely applied to the field of commodity consumption; and compared with the prior art, different ages, different genders and other features can be distinguished, so that the cloud computing application is widened, people obtain better experience in a consumption pattern, and knowing the development direction of a product and the consumption demand of a consumer to establish a good interaction between the consumer and the company is facilitated for the company.

A method and apparatus for computing correlating images for image recognition is provided. In particular, the method is used to compare images of a biometric object to authorize an individual or verify whether an individual is the person he claims to be. The biometric objects may include fingerprints, hand or palm prints, and retina scans. The method uses a deviation of symmetry of correlation functions to determine whether the images are of a same object.

A method provides for encoding a frame of video using an edge map made up of one or more edge-blocks detected in the frame. When the edge-blocks are contiguous, at least one slice partition is formed using the edge-blocks and the slice partition is encoded using a sample adaptive offset (SAO) filter, wherein the slice partition is formed with an adaptive slice width, and the sample adaptive offset (SAO) filter is turned on or off during the encoding based on whether the edge-blocks are being encoded. When the edge-blocks are not contiguous, edge-block processing is performed around edges in the frame during encoding of the edge-blocks. The edge-block processing involves configuring one or more of: an intra block size, a transform block size, an inter prediction block size, a quantization parameter, candidate modes for intra prediction, pyramid level for motion estimation, and fractional pixel motion estimation search.

A method and tool for conducting charged-particle beam direct write lithography is disclosed. A disclosed method involves condensing an initial design file down to a set of profiles and a pattern of relative locations to form a formatted pattern file. The formatted pattern file is adjusted to accommodate desired pattern corrections. Portions of the formatted pattern records are extracted to form data strips that have a plurality of channels with a pattern of profiles and spatial indicators. Data strips are sequentially read to construct a printable pattern of profiles and spatial indicators that specify the locations of the profiles. Additionally, the pattern of profiles are sequentially printed from each data strip onto a substrate to form the desired pattern on the substrate.

The invention provides a method and a device for an integrating two-dimensional code with an image. The method of the invention comprises steps of adjusting the attributes of part of points in the two-dimensional code to enable the attributes of part of the points to be close to the attributes of the corresponding areas in the first image according to attribute information of a first image and attribute information of the points in the two-dimensional code of the first image, wherein the two-dimensional identification result of color lumps to which part of the points the attributes of which are not adjusted belong to is different from the two-dimensional recognition result of the area in the first image corresponding to the part of the points, and integrating the two-dimensional code with the first image to obtain a second image, wherein the attributes of points of the two-dimensional code are adjusted. The invention enables the two-dimensional code to be more matched with the environment color.

A method, medium, and system effectively compressing and/or restoring binary images. By compressing pixel values making up a 2×2 block in an input image, using representative values representing the pixel values and a pattern of the pixel values represented by the representative values, it is possible to effectively compress and/or restore binary images having little similarities between pixel values. In addition, by extracting compressed values and a pattern of pixel values making up a 2×2 block in an image, from compressed data of the 2×2 block, and restoring pixel values using the compressed values and the pattern, it is possible to effectively restore binary images.

The invention discloses a multi-mode collaborative learning method and apparatus based on a neural network. The method comprises the following steps: S1 capturing short video data to establish a shortvideo data set, wherein the short video data comprise short videos published by a user historically and subsidiary information thereof; S2 extracting multi-mode features of the short video data in the short video data set, and performing information complement for the multi-mode features; S3 sequentially taking any modal feature in the modal features as a main mode and using the modal features asauxiliary modes, calculating the relationship weights of the main mode and the auxiliary modes, respectively classifying identical feature information and supplementary feature information, and performing information recombination on the supplementary feature information and the identical feature information enhanced by the neural network; and S4 establishing an attention mechanism network, distinguishing the degree of contribution of each piece of recombined mode feature information a scene category to which the short videos belong, and performing short video scene classification by using the neural network.

A method of detecting a target in a scene comprises the steps of generating a comparison of the value of data elements in first and second data sets, the data elements corresponding to returns from the same part of the scene and setting a detection threshold value for the comparison so as to detect targets entering said that part of scene The method provides improved target detection in the presence of clutter.

An image processing apparatus includes a corrector for correcting a pixel value of each of a plurality of pixels forming each access unit to be processed, based on at least one of the parameter values corresponding to the access unit to be processed. The corrector includes an acquisition unit for acquiring input pixel values, a first mean value calculating unit for calculating a first mean pixel value, a second mean value calculating unit for calculating a second mean pixel value, a correction value determination unit for determining a correction value, and an adding unit for adding the correction value to the input pixel value of a target pixel.

The embodiment of the invention discloses an eye fundus image recognition method and device, electronic equipment and a storage medium. The method comprises the steps of obtaining a to-be-identified fundus image, and performing grid partitioning processing on the to-be-identified fundus image to form a plurality of to-be-identified partitioned grids; Inputting each to-be-identified block grid of the to-be-identified fundus image into a pre-trained neural network model to determine a focus state of the to-be-identified block grid; And determining the focus state of the to-be-identified fundus image according to the focus state and the position of each to-be-identified block grid in the to-be-identified fundus image. According to the technical scheme provided by the embodiment of the invention, the number of required fundus image samples and the data calculation amount in the model training process are reduced while the focus recognition accuracy and multi-type focus recognition are ensured.

The invention discloses an emotion recognition method and system based on a deep learning model and a long-short memory network, The method comprises the following steps: data preprocessing and data set partitioning of EEG signals are performed to construct a network model, wherein the network model comprises a picture reconstruction model composed of a variational encoder and an emotion recognition model composed of a long-short memory network; the network model comprises an image reconstruction model composed of a variational encoder and a short-long memory network; The objective function isconstructed according to the network model. The network model is trained by training set, and the objective function is optimized by Adam optimizer in neural network, and the trained network model isobtained. Using the cross-test set to cross-test the trained network model, determining the super-parameters of the network model, and obtaining the final network model; and using the final network model to visualize the seed data and perform emotion recognition. The invention relies on the data artificial intelligence method to learn the collected EEG signal space and time complex structure, reduce the subjective factors in the prediction, and improve the prediction accuracy.

The invention relates to a cloud point filtering method and device, computer equipment and a storage medium. The method comprises that cloud point data is obtained; the cloud point data includes a setof data points obtained from the road scanning environment; a neighborhood of a present data point in the point cloud data is determined; when the counted first number of data points in the neighborhood of the present data point is lower than a preset threshold, the present data point is determined to be a candidate noise point; the candidate noise point as a seed point carries out regional growth to obtain a target growth region; when the counted second number of data points in the target growth region is lower than a preset threshold, the candidate noise point as a noise point is filtered from the point cloud data. The noise point is located preliminarily, and then determined by regional growth and filtered, data points formed by objects in the far place and sparse data points in the point cloud data are protected, and the accuracy of point cloud filtering is improved.

A method of determining input pattern is adapted to be implemented on an electronic apparatus equipped with a touch panel and includes the steps of: detecting a plurality of boundary points between an input pattern inputted through the touch panel and a circumscribed polygon of the input pattern, detecting an area ratio of a polygon defined by the boundary points to the circumscribed polygon, and determining the shape of the input pattern at least according to the area ratio. The present invention also provides a computer readable storage medium having a program stored therein. When the program is executed which enables an electronic apparatus equipped with a touch panel to determine the shape and/or direction of an input pattern inputted through the touch panel.

A method to transmit a label between two images, characterized in that the method includes the following steps:

• • providing a first image, the first image comprising several sets of connected points, each set being characterized by a label,
  • providing a second image,
  • from the second image, determining several sets of connected points,
  • superimposing the two images to determine the common parts and non-common parts of the sets in the first and second image,
  • giving each common part of the second image, the label of the set in the first image with which said part is common,
  • giving each non-common part of the second image in contact with a single set of connected points in the first image, the label of said set,
  • giving a new label to each non-common part of the second image not in contact with any set in the first image.

The present invention provides a face identification method and device. The method comprises: obtaining a user' face image; performing key point coordinate detection of the face image, and employing the detected key point coordinates to perform correction of the face image; recalculating the key point coordinates for the corrected face image, and extracting user' face features according to the recalculated key point coordinates; and employing the face features and the recalculated key point coordinates to perform face verification. Therefore, key point coordinates of a face image are detected,the face image is corrected through adoption of the key point coordinates, the corrected face image is subjected to extraction of user' face features, and face verification is performed, so that precision and accuracy of face identification are improved.

The present invention relates to a super-resolution image quality evaluation method. The method comprises: employing a pixel corresponding relation between a low-resolution image and a super-resolution image to obtain a corresponding relation between local image blocks, respectively solving the energy and the texture information of the low-resolution image and the super-resolution image, and obtaining the vision similarity between the changing of the two features. The final super-resolution image quality is obtained through evaluation of the vision similarity of the local image blocks of the low-resolution image and the super-resolution image. The experiment result shows good effects in the quality evaluation of the super-resolution image.

The invention provides a hand-held terminal with the handwriting recognition and identity validation function and a realizing method, which belongs to the electronic equipment field and relates to the hand-held terminal which is realized via handwriting recognition and the assorted realizing method. The terminal comprises a handwriting input structure, a handwriting recognition module and a handwriting recognition authority management module. The users can input handwriting of own in the hand-held terminal by utilizing the handwriting input structure in the invention, and can effectively be processed identity validation by the handwriting recognition module, thereby the authority of the users to enter the hand-held terminal can be determined; the users need not clearly remember the password, only need handwriting recognition to process authority recognition, and in the state of guaranteeing privacy and safety of the hand-held terminal, a great convenience is brought for the users.

The invention discloses a video-based vehicle tracking method and device. The method comprises: determining a first image frame for tracking a vehicle in a video and using the first image frame to initialize a model; when processing a current image frame, using a determined searching range in a previous image frame to extract image features within the searching range in the current image frame, generates a feature cycle matrix, matching the feature cycle matrix with the current model, and obtaining a matching degree matrix; updating a target position according to a matching degree value of each position of the matching degree matrix to track the vehicle; before generating the model of each image frame, redefining the target size based on the calibration, and zooming the searching range of the previous image frame according to the proportional relation between the target size and the searching range; extracting image features in the searching range in the current image frame, finally generating a model according to the image features, and updating the current model. As the target size of each image frame is updated according to the calibration, the system can track the target in adaptive scales, so that the vehicle tracking is more accurate.

The invention discloses a stainless steel weld defect efficient detection method based on multi-domain expression data enhancement and model self-optimization. The method comprises the following steps: deriving a one-dimensional echo time domain signal to spatial domains such as a time-frequency domain, a Gramb angle field domain and a Markov transfer field domain; sequentially inputting the data set constructed by each spatial domain into the MobileNetV3 neural network, and selecting the spatial domain with the most abundant feature expression as a final training data set; constructing a multi-scale depth separable convolution to improve the MobileNetV3 so as to enhance the recognition performance of the network; providing a particle swarm-chaos sparrow search algorithm for automatic optimization of a network structure and parameters; and adopting the CPU + FPGA heterogeneous cooperative calculation to accelerate the reasoning and recognition application speed of the defects. Five types of weld defects such as incomplete fusion, air holes, slag inclusion, incomplete penetration and cracks are taken as objects, the recognition accuracy of the five types of weld defects can reach 98.75%, and the method has practical engineering application value.